Multilingual PDFs — Font Embedding Pitfalls

The proposal PDF mixed English, Urdu, and simplified Chinese. On the client's Windows laptop, Urdu became hollow rectangles. Chinese survived — they'd embedded Noto for CJK but assumed "Arial Unicode" would cover Nastaliq.

Multilingual PDFs fail quietly. Embedding isn't automatic even when it looks fine on your Mac.

How PDF fonts work

PDF stores text as glyph references plus optional embedded font programs. Viewers render using:

1. Embedded font (reliable)
2. System substitution (unreliable across scripts)
3. Failure → tofu □□□

Scripts with highest risk

| Script | Common pitfall | | --- | --- | | Arabic / Urdu | Joining forms need correct OpenType features | | Devanagari (Hindi) | Missing matras with wrong font | | CJK | GB vs Big5 legacy encodings | | Thai | Combining marks spacing | | Emoji | Color glyph layers dropped |

One font rarely covers everything — stack fonts per script in source document.

Embedding checklist

• [ ] Export setting: Embed all fonts / ISO 19005 PDF/A if archival
• [ ] License allows embedding (some desktop fonts forbid it)
• [ ] Subset includes all diacritics used
• [ ] Bold/italic are separate embedded faces, not faux-bold
• [ ] No "standard 14" PDF fonts for non-Latin — Times won't save Urdu

Deep dive: Embedding Fonts When Exporting Word to PDF.

Word, InDesign, Google Docs

Word — Options → Save → embed fonts in file; export PDF with best for printing.

InDesign — Preflight embed fonts; export PDF/X standards.

Google Docs — Limited font list; complex scripts may need Word/InDesign finish.

Use Word to PDF and verify output.

Web fonts in generated PDFs

Headless Chrome PDF from HTML — ensure @font-face fonts load before print. Missing wait = fallback fonts in PDF.

File size trade-offs

CJK full subsets add megabytes. Compress PDF after verifying glyphs — OCR Before Compression for scanned multilingual docs.

Accessibility

Tagged PDF with lang attributes per block — PDF Accessibility Basics. Screen readers pick correct voice per language.

Legal and government filings

Some courts require PDF/A with embedded fonts only. Test with preflight tools before upload.

Legal PDF Page Numbering unrelated to fonts but same filing anxiety — test everything on filing portal preview.

Student submissions

Student Assignment PDF — bilingual essays need embedding proof.

Troubleshooting

Why do some characters show as boxes in PDF? The font used for those characters wasn't embedded, and the viewer substituted a font that lacks the glyphs. Common with Arabic, Hindi, Chinese, or rare Unicode symbols when export settings skip embedding.

What does embed fonts mean in PDF? The PDF file includes font data (full or subset) so any device renders text identically without relying on locally installed fonts.

Should I embed full fonts or subset? Subsetting embeds only characters used — smaller file size, standard for most exports. Full embedding needed rarely — some licensing or editing workflows. Both fix missing glyph issues if subset includes all used codepoints.

Limitations

When not to use this approach

Conclusion

Embed per script with licensed fonts. Test on a machine that isn't yours.

Tofu boxes mean someone assumed your fonts travel with the file — they don't unless embedded. Open Urdu on Windows before the client does.

Right-to-left (RTL) layout

Arabic and Urdu need RTL paragraph direction in source document — embedding font alone doesn't fix LTR layout with broken glyph joining. Test bidirectional text with numbers and Latin inserts (common in invoices).

Font subsetting size management

CJK documents may reach 20+ MB from subsetting alone — consider splitting by chapter or language section. Compress PDF after subset verification.

Google Noto family strategy

Noto Sans + Noto Nastaliq Urdu + Noto Sans CJK as stack covers many scripts — verify licenses permit embedding in commercial PDFs distributed to clients.